This is an excellent, if a bit wonky, article on how AIchat does what it does.
..
purpose here is to give a rough outline of what’s going on inside ChatGPT—and then to explore why
it is that it can do so well in producing what we might consider to be meaningful text
Stephen Wolfram explores the broader picture of what's going on inside ChatGPT and why it produces meaningful text. Discusses models, training neural nets, embeddings, tokens, transformers, language syntax.
writings.stephenwolfram.com
And he is just covering part of the architecture. As good as that explanation is it doesn't cover the "Reinforcement learning" part.
We’ve trained a model called ChatGPT which interacts in a conversational way. The dialogue format makes it possible for ChatGPT to answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests. ChatGPT is a sibling model to InstructGPT, which is...
openai.com
Methods
We trained this model using Reinforcement Learning from Human Feedback (RLHF), using the same methods as
InstructGPT, but with slight differences in the data collection setup. We trained an initial model using supervised fine-tuning: human AI trainers provided conversations in which they played both sides—the user and an AI assistant. We gave the trainers access to model-written suggestions to help them compose their responses. We mixed this new dialogue dataset with the InstructGPT dataset, which we transformed into a dialogue format.
To create a reward model for reinforcement learning, we needed to collect comparison data, which consisted of two or more model responses ranked by quality. To collect this data, we took conversations that AI trainers had with the chatbot. We randomly selected a model-written message, sampled several alternative completions, and had AI trainers rank them. Using these reward models, we can fine-tune the model using
Proximal Policy Optimization. We performed several iterations of this process.
